METAL LAB
Add us on GoogleMETAL

Claude Code Gets Faster Startup, Clearer Token Tracking, Remote Control Fixes

Anthropic rolled out updates to Claude Code this week, speeding up CLI startup, adding token usage visibility, and making auto-mode rules editable

이미지: METAL LAB 생성

Summary

  • Anthropic's Claude Code shipped an update this week that speeds up CLI startup and shows more detailed token usage
  • Users can now see and edit exactly which rules apply in auto mode, and additional bugs in the remote control feature were fixed
  • The official account said the changes reflect user feedback, but didn't share specific figures or a rollout timeline

What changed in this week's update

Anthropic official website

A user drawn as a scattered cluster of dots and Claude Code drawn as an orbiting satellite are connected by a dotted, bidirectional arrow. The user sends feedback about friction points, and Claude Code responds with small weekly fixes, repeating the orbit.A user drawn as a scattered cluster of dots and Claude Code drawn as an orbiting satellite are connected by a dotted, bidirectional arrow. The user sends feedback about friction points, and Claude Code responds with small weekly fixes, repeating the orbit.

ClaudeDevs, the account covering Anthropic's coding tool Claude Code, posted on August 28 about four improvements rolled out this week. The team sped up CLI startup — the mode where users control the program through typed commands only — and made it easier to see where tokens are being spent during a task. They also made it simpler to view and edit the rules applied under auto mode, and fixed additional bugs in the Remote Control feature, which lets users manage tasks remotely. The account offered only a brief explanation, saying all of these changes reflect user feedback, without disclosing the specific size of the speed improvement or which platforms it applies to.

Why Claude Code keeps getting these frequent tweaks

Claude Code is Anthropic's coding agent, letting people write and fix code by typing commands straight into a terminal. It's seen frequent changes over the past several weeks. The pattern suggests the Claude Code team is favoring a steady stream of small fixes — addressing slow response times, opaque costs, confusing automation rules — over dropping one big feature at a time. Because a coding agent edits files and runs commands even when no one's watching directly, being transparent about what's actually happening becomes central to earning trust. Making token usage more visible and letting users edit auto-mode rules both fit that same logic.

68309ab48369f7ad9b4a40e1 open graph

So what does this actually mean for users

For developers who keep Claude Code running all day and hand it long tasks, the faster startup should ease the friction of opening and closing sessions, and the more detailed token usage breakdown should make cost tracking easier. Teams that hand off permissions to auto mode can now check exactly which rules are in effect and adjust them right away if needed. Anyone relying on Remote Control for tasks run from a distance should see a bit more stability from this round of fixes. That said, the announcement didn't include specific performance numbers or a version number, so how much difference users actually notice will depend on trying it out firsthand.

Comments

Related stories